The Amsterdam-listed group is putting conversational app-building in front of restaurants and shopkeepers, the users it says AI has so far left behind.

Prosus has launched ToqanClaw, a platform that lets people build apps, dashboards, and automations by describing what they want in plain language, the way they would explain a task to a colleague.

The Amsterdam-listed technology group unveiled it today, billing itself as the first company in Europe to put OpenClaw-style tools in front of its business partners at scale.

The idea is to remove the engineer from the loop. A restaurant owner who wants a delivery-analytics dashboard, or a shopkeeper who wants to automate a weekly report, can describe the tool in a conversation and have it built without writing code, opening a ticket, or waiting on an IT team.

Prosus says the platform is ready immediately and that it routes across more than 20 underlying AI models to pick the best one for a given task, which it claims makes it cheaper than the alternatives.

ToqanClaw is built on Toqan, the in-house AI platform Prosus has been developing for its own staff, and it inherits that system’s data posture: the company says a customer’s data stays under their control and is never used to train third-party models.

That detail is the point of the whole exercise. Earlier reporting had described Prosus building an OpenClaw rival to sidestep European privacy concerns, and ToqanClaw is the product that reporting was pointing at.

It is a different posture from rivals that have built directly on the platform, as Tencent did with its ClawPro enterprise agents.

The target users are the ones Prosus argues conventional AI has skipped. It is making the platform available to more than five million restaurants, merchants, and entrepreneurs across its ecosystem, businesses that have lacked the technical teams to build software of their own.

The framing leans on the same democratisation argument that has surrounded the wider rise of OpenClaw and its agent tools, now aimed at small merchants rather than developers.

Prosus put numbers behind its own use of the approach. Chief executive Fabricio Bloisi said the group had spent 18 months building internally, reaching 60,000 agents and 10,000 applications made by people who had never written a line of code.

The company also said it has trained a specialised commerce model, which it calls the Large Commerce Model, on data from more than a billion customers and 500 million daily interactions, and that connecting it to ToqanClaw lets agents start anticipating what a business needs rather than only executing instructions.

The launch came with customer examples, all company-supplied. Lebkov & Sons, a Dutch cafe chain, is said to have cut financial reporting from weeks to 30 minutes.

Burger & Frites, a Rotterdam burger chain, built a delivery-analytics agent that the company says saves it about €21,000 a month; and Poke Perfect, a poke-bowl chain, made a WhatsApp-based operations assistant that reduced routine staff queries by 70%.

The figures are Prosus’s, not independently audited.

Alongside ToqanClaw, Prosus also pushed Zapia, a consumer-facing AI assistant it backs, into wider release.

Zapia handles tasks end to end, the example given being to find a restaurant, shortlist options in a family group chat, wait for votes, and book the table, and Prosus says more than six million people already use it, mostly in Latin America.

It is now available on the App Store, Google Play, and the web, with a free tier covering most personal use and a paid plan for heavier users.

Both launches were timed to Prosus Forward, the group’s inaugural product event, where it set out an AI-first pitch built around the same idea: that owning the data and the customer relationship, rather than the model, is the advantage worth having.

The harder test, as with any tool that promises to build software from a sentence, is whether merchants keep using what they make once the demo is over.